Journal of Economic Entomology Advance Access published May 26, 2015 ECOLOGY
AND
BEHAVIOR
Regional-Scale Spatio-Temporal Analysis of Anastrepha ludens (Diptera: Tephritidae) Populations in the Citrus Region of Santa Engracia, Tamaulipas, Mexico VENANCIO VANOYE-ELIGIO,1,2 LUDIVINA BARRIENTOS-LOZANO,3 ˜ EDA,4 GRISELDA GAONA-GARCI´A,1 AND ROBERTO PE´REZ-CASTAN MANUEL LARA-VILLALON1
J. Econ. Entomol. 1–10 (2015); DOI: 10.1093/jee/tov134
ABSTRACT Large citrus areas in Tamaulipas are affected by Anastrepha ludens (Loew) populations. Here we report the findings of a spatio-temporal analysis of A. ludens on an extended citrus area from 2008–2011 aimed at analyzing the probabilities of A. ludens infestation and developing an infestation risk classification for citrus production. A Geographic Information System combined with the indicator kriging geostatistics technique was used to assess A. ludens adult densities in the spring and fall. During the spring, our models predicted higher probabilities of infestation in the western region, close to the Sierra Madre Oriental, than in the east. Although a patchy distribution of probabilities was observed in the fall, there was a trend toward higher probabilities of infestation in the west than east. The final raster models summarized the probability maps using a three-tiered infestation risk classification (low-, medium-, and high risk). These models confirmed the greater infestation risk in the west in both seasons. These risk classification data support arguments for the use of the sterile insect technique and biological control in this extended citrus area and will have practical implications for the area-wide integrated pest management carried out by the National Program Against Fruit Flies in Tamaulipas, Mexico. KEY WORDS Mexican fruit fly, distribution, indicator kriging, risk, GIS
Citrus growing is an important economic activity in northeastern Mexico. Tamaulipas has about 38,000 ha of citrus areas (Secretarı´a de Agricultura, Ganaderı´a, Desarrollo Rural, Pesca y Alimentacio´n [SAGARPA] 2013), making it the third largest Mexican citrus producer after the states of Veracruz and San Luis Potosı´. Important citrus-growing areas are organized in large blocks that are subject to area-wide integrated pest management (AW-IPM) aimed at suppressing fruit fly (Tephritidae) populations (Hendrichs et al. 2007; Gutie´rrez 2010). Anastrepha ludens (Loew) (Diptera: Tephritidae), also known as the Mexican fruit fly, is one of the most important citrus pests affecting areas from the United States to Costa Rica (Hernandez-Ortiz and Aluja 1993). A. ludens is considered native to the Sierra Madre Oriental in northeastern Mexico because of the presence of its wild host Casimiroa greggi (S. Watson) Chiang (Plummer et al. 1941), known as yellow chapote, which
1 Instituto de Ecologı´a Aplicada, Universidad Auto´noma de Tamaulipas, Divisio´n del Golfo 356, Colonia Libertad, C.P. 87019, Cd. Victoria, Tamaulipas, Me´xico. 2 Corresponding author, e-mail:
[email protected]. 3 Instituto Tecnolo´gico de Ciudad Victoria, Boulevard Emilio Portes Gil No. 1301, Cd. Victoria, C.P. 87010, Tamaulipas, Mexico. 4 Facultad de Medicina Veterinaria y Zootecnia, Universidad Auto´noma de Tamaulipas, Carretera Victoria-Mante Km 5, A.P. 263, Cd. Victoria C.P. 87000, Tamaulipas, Me´xico.
can present higher rates of damage than the commercial hosts in Tamaulipas (Mangan et al. 1997). The Mexican National Program Against Fruit Flies (NPAFF) aims to control, suppress, and eradicate fruit flies in fruit growing (Gutie´rrez 2010, Suckling et al. 2014). In Tamaulipas, the NPAFF began in 1994 in backyards, native sites, and citrus areas and is mainly applied to Valencia orange (Citrus sinensis (L.) Osbeck), grapefruit (Citrus paradisi Macfad), and mandarin (Citrus reticulata Blanco). These citrus varieties are particularly susceptible to infestation during the ripening in spring and in the fall-winter (Thomas 2003, Birke et al. 2006). The NPAFF has been monitoring activities in a McPhail (Steyskal 1977) trapping network in major citrus areas of Tamaulipas (e.g., the municipalities of Hidalgo, Padilla, Gu¨emez, Victoria, and Llera) since 1994. The introduction of a Geographic Information System (GIS) and a Global Positioning System (Liebhold and Sharov 1998, Scalon et al. 2011) have enabled the spatial distribution analysis of A. ludens captured adults across large citrus areas. However, this approach considers only current and point data and applies economic thresholds used to recommend bait sprays. Although this approach is usually effective in the short-term, it has little or no impact on future populations, particularly in areas with high probabilities of infestation. The installed trapping network has revealed a spatial heterogeneity of A. ludens adult densities within the
C The Authors 2015. Published by Oxford University Press on behalf of Entomological Society of America. V
All rights reserved. For Permissions, please email:
[email protected]
2
JOURNAL OF ECONOMIC ENTOMOLOGY
citrus areas. This suggests a combination of human and ecological factors favoring survival of populations in certain sites, which results in distinct patches with varying degrees of isolation (Tscharntke et al. 2002). The spatial variation of pest densities in these areas could be used to establish different risk levels and define homogeneous zones, which might help optimize pest monitoring and reduce overall costs (Fleischer et al. 1999, Castrignano` et al. 2012). Spatial analyses in Tephritidae have revealed a heterogeneous population distribution at distinct scales, with focus on Ceratitis capitata Wiedemman (Israely et al 2005, Alemany et al. 2006, Epsky et al. 2010, Lira 2010), Bactrocera oleae Rossi (Castrignano` et al. 2012), Bactrocera tryoni (Froggatt), and Bactrocera papayae (Drew and Hancock) (Meats 2007). Previous local-scale studies of A. ludens and C. capitata have investigated effectiveness of sterile fly dispersion (Baker and Chan 1991, Plant and Cunningham 1991, Thomas and Loera-Gallardo 1998). Furthermore, the spatial component in Tephritidae plays a decisive role in a spatial decision support system (Cohen et al. 2008). Geostatistics are increasingly being used in many aspects of insect ecology and pest management (Liebhold et al. 1993, Brenner et al. 1998, Fleischer et al. 1999). The number of traps and their geographic locations are used to determine basic spatial patterns in population predictions and to identify sources of infestation (Epsky et al. 2010), as well as the distribution of insect populations and infestations of sites not sampled (Guidotti et al. 2005, Garcı´a 2006). The geostatistical technique, known as the indicator kriging (IK), is a form of ordinary kriging based on the semivariogram. The indicator semivariogram function treats data as binary indicators with respect to a threshold value and thus makes it possible to capture extreme values (Journel 1983). IK focuses on performing probability models from categorical data that may be useful in decision making in precision pest integrated management (Fleischer et al. 1999). The complete methodology of the IK approach is addressed in Goovaerts (1997) and Deutsch and Journel (1992). Based on the above, the geographical conditions of the citrus growing in Tamaulipas allow the use of infestation level as a variable in a geostatistical approach into a GIS for mapping and evaluation within a delimited space. Here we aimed to analyze the probabilities of infestation of A. ludens populations in a large citrus area and to determine an infestation risk classification in the main seasons of citrus production. To our knowledge, this is the first attempt to study the spatial distribution of A. ludens populations on a regional scale in northeastern Mexico. Materials and Methods Study Site. The citrus region of Santa Engracia is centrally located within the state of Tamaulipas. Santa Engracia comprises 8,846 ha including urban, marginal, and citrus areas, between N 24 03’ and 23 57’ and W 99 19’ and 99 07’, 30 km north of the capital city (Ciudad Victoria) and forms part of the
municipalities of Hidalgo and Gu¨emez (Fig. 1). The temperature ranges between 5 (winter) and 40 C (summer), and rainfall data values ranged between 78 and 130 mm per year during the study period. The altitude varies from 200 to 300 meters above sea level and the region has a flat topography, bordered to the west by the Sierra Madre Oriental and with the Corona River crossing the region. Hosts and Monitoring. The Santa Engracia region has 6,500 ha of citrus crops, consisting mainly of Valencia orange, grapefruit, and mandarin (SAGARPA 2013). Backyards tree fruits are also present in the study area, particularly sour orange (Citrus aurantium L.). The study was based upon the number of weekly captured flies in 181 McPhail traps deployed and supervised by the NPAFF, between 2008 and 2011. Each trap was georeferenced in a Universal Transverse Mercator (UTM) coordinate system (14 N zone and WGS 84 datum) and projected in a shapefile format (ESRI Inc. 1998). McPhail traps were baited with four tablets (20 g) of torula yeast (Trampol 300, Grupo PAUSA, S.A. de C.V. Mexico, D.F.) mixed with 250–300 ml of water and hung in a tree between 3 and 4 m above the ground. Traps were serviced every 7 d and placed as uniformly as possible in the entire region. Host availability was the main criterion for placing traps, although in large citrus areas different sweet varieties are commonly intercalated inside orchards. Weekly trap captures were converted to flies per trap per day (FTD) values (International Atomic Energy Agency [IAEA] 2003) using the following formula: FTD ¼
F T x D
where F is the number of captures flies, T is trap number, and D is the number of days of trap exposure. This way, the variability in trap number and adults captures was standardized (Aluja et al. 2012). Geostatistical Analysis. Two seasons of 17 wk were considered for every year of the study period. The Valencia orange season is first amongst the considered citruses and covers the period of January to April (spring). This is the most important citrus production season in Tamaulipas. An average FTD value for the 17-wk period was calculated for every trap. The second season (September to December [fall]) coincides with the ripening of early citrus varieties such as grapefruit, mandarin, and early orange. The average FTD of each trap were also calculated for this 17-wk period. In this way, the average FTD for every trap in each season (spring and fall) represents the variable for the geostatistical study. Additionally, a general average FTD of all traps was evaluated using descriptive statistics for the two seasons per year. The latter data were analyzed without considering their geographical distribution. The FTD unit of measurement is a recognized international reference value (IAEA 2003). Therefore, to perform IK, the setting of FTD thresholds was based
2015
VANOYE-ELIGIO ET AL.: SPATIAL ANALYSIS OF A. ludens IN MEXICO
3
Fig. 1. Geographic location of the citrus region of Santa Engracia region, Tamaulipas, Mexico. Spatial distribution of McPhail traps from 2008–2011.
on the availability of commercial citrus hosts and fruit infestation levels, which are generally higher in the first months of the year. Indicator values ðIi Þ were calculated by setting an FTD threshold ðzc ) and assigning a value of 1 to McPhail traps (sampling locations) ðzi Þ with an FTD of less than or equal to the FTD threshold and a value of 0 to traps where the FTD was greater than the FTD threshold. Thus, indicator values of FTD for sampling locations were defined by: Ii ðzc Þ ¼ 1 if zi zc or 0 if zi > zc
(1)
where Ii ðzc Þ is the indicator value at location i, zi is the FTD at location i, and zc is the FTD threshold. Indicator semivariograms per year were performed using a zc (FTD) of 0.1430 in spring and 0.0500 in the fall, based on the indicator values defined by equation (1) using the sample semivariance (Lyon et al. 2006), cs ðhÞ, at a lag, h, of cs ðhÞ ¼
1 2NðhÞ
X
ðIi ðzc Þ Ij ðzc Þ Ij ðzc ÞÞ2
(2)
ði;jÞ
where N is the number of pairs, Ii ðzc Þ and Ij (zc ) are the indicator values at i and j, respectively, with
summation over pairs (i; jÞ for the lag bin. This varied from standard semivariance calculations by using indicator values at points i and j instead of measured values (Lyon et al. 2006). A correct modeling of local uncertainty is discussed in Barabas et al. (2001) and Grunwald et al. (2006). Probability models were performed for the identification of areas with different infestation risk in each season per year. These models were reclassified into four probability categories of exceeding settled thresholds: 0–25, 25–50, 50–75, and 75–100%. The reclassification is a generalization process used to reassign values in a raster input layer such as the position, value, shape, size, cell measure, the level of contiguity and thereby create a new data layer (Eydel et al. 2011). For the reclassification, Reclassify tool of the module Spatial Analyst of the software ArcGis 9.3 (ESRI Inc. 2008) was used. To perform IK, Geostatistical Analyst module in ArcGis 9.3 was used, which determined the best experimental and theoretical indicator semivariogram fitted to the points and the presence of some directional pattern of the populations (anisotropy). For indicator semivariograms three parameters were evaluated: nugget effect, a discontinuity in the origin associated with smaller spatial variation than the shortest distance in
JOURNAL OF ECONOMIC ENTOMOLOGY
4
the sampling interval (Goovaerts 1997). The sill that considers the sample variance and range that indicates the maximum distance within which there is a relation between pairs of data (Barabas et al. 2001). A variable like the insect density is distributed very erratically in reduced distances, and exponential or spherical models are the most suitable (Isaaks and Srivastava 1989). Risk Classification. Probabilistic models were transformed to raster format to perform the final risk model of each season. In its simplest form, a raster consists of a matrix of cells (or pixels) organized into rows and columns (or a grid) where each cell contains a value representing information, in this case probability value (0–1). The cell size was determined on 10,000 m2 (1 ha) and each probability value between 0 and 1 of exceeding settled threshold was represented. Overlay maps require local operations with two or more layers where each raster of each layer has the same geographical location and thus a new layer ðSÞ containing the combination of the information of the input raster is generated (Eydel et al. 2011). The local operation with the raster layers of each season of the study is defined by equation (3). S¼
X xi n
(3)
where S is the risk final model, xi represents an input raster, and n is the number of the input raster. Then, a new raster layer was performed and a reclassification process was carried out with the following categories: low-risk between 0–30%; medium-risk between 30–60%; and high-risk 60–100%. In this step, the ArcGis 9.3 Spatial Analysis module, Reclassify, and Raster Calculator tools were used for local operations in the raster layers. Results The population fluctuations of A. ludens showed a clear delimitation of two seasons of major infestations in the region of Santa Engracia. Population peaks of Mexican fruit fly exhibited a consistent temporal dynamic throughout the study. Furthermore, we observed a significant reduction in the number of flies captured in the months of May to August. The scarcity of citrus production was the major factor of population decline in these months (Fig. 2). Descriptive Statistics. Analysis of the general average FTD revealed higher mean values of the FTD above median values, indicating positive skewness in both seasons (Tables 1 and 2). A high skewness and mean is always larger than the median were indications of a lognormal distribution which was observed during the study. Geostatistical Analysis. The spatial continuity of the A. ludens population density was described by an exponential model (theoretical semivariogram) in the experimental semivariograms, which provided the best fit for all years (Tables 3 and 4). The spatial autocorrelation of FTD showed spatial dependence among the traps (range) that extended over short distances. This
distance varied from 400–850 m in spring and 800– 900 m in fall (Tables 3 and 4). Semivariograms showed variations in the nugget effect for the two seasons. This variation was probably caused by the annual difference in the population density of the pest and the trapping scale, which was independent of population density. The ratio between the nugget effect and sill was greater than 50%, assuming a consistent definition in the models (Tables 3 and 4). For spring 2009, the probability models noted a predominance of areas with infestation probabilities of between 75 and 100%, exceeding the FTD of 0.1430. However, even in this context, a trend toward a higher percentage was seen in the center-west. Moreover, probabilities of between 25–50% were plotted in the east (Fig. 3). In 2008, the density of areas with a high probability of infestation increased from east to west; a homogeneous area of between 0 and 25% risk surrounded by areas of 25–50% risk could be clearly defined. These findings were likely due to the consistently high FTD levels for those years (Fig. 2). In contrast, the nature of the variation shifted significantly during 2010–2011, with less consistently high FTD levels than for 2008 and 2009. In these years, larger areas showed probabilities from 0–25% and the east to west trend was sustained. Despite the low populations in 2010 and 2011 a patch formation associated with high FTD levels could be observed in the west. These sites could be assumed to be high-risk areas or foci of infestation. For the fall of 2008, probability models identified an area that mostly ranged from 75–100% probability and exceeding the FTD threshold of 0.0500. Only a few small patches of 0–50% probability were detected in the east. In 2009, the FTD was lower and a heterogeneous distribution of probabilities was observed, with the high probability areas being in the west and center. In 2010, a decline in FTD promoted an increase in the proportion of the citrus area within the 0–25% probability range. Patches in the center and southwestern region with high probabilities (50–75%) could be recognized; although high probabilities were typically embedded in low probability areas (0–25%). In 2011, although low probability sites prevailed, some areas with probabilities higher than 50% were observed at the center and southwest of the region (Fig. 4). Risk Classification. The surface of the final raster covered 8,846 ha and included the geographic region of Santa Engracia (Table 5). A probability distribution was clearly observed in both seasons of citrus production. The raster layers summarized the results from the probability models into risk levels. These raster models again highlight the tendency for higher risk in the western region (in both spring and fall), although this trend was more evident than in the fall. The high-risk surface accounted for 10.8 and 1.7% of the total cell of the raster models in spring and fall, respectively. The majority of the surface (73.9% in spring and 65% in fall) was characterized by a medium-risk level in both seasons, which can be considered as a transitional process from low to high-risk (Table 5). Low-risk sites were mostly observed to the
2015
VANOYE-ELIGIO ET AL.: SPATIAL ANALYSIS OF A. ludens IN MEXICO
5
Fig. 2. Weekly population fluctuation (FTD) of A. ludens at Santa Engracia, Tamaulipas, 2008–2011. Study seasons (spring and fall) according to the citrus production and abundance of Mexican fruit fly captures. Table 1. Descriptive statistics of general average FTD of A. ludens in 17 wk 2008–2011 in spring at Santa Engracia, Tamaulipas Year
Mean
Median
Variance
2008 2009 2010 2011
0.1944 0.3689 0.1132 0.1295
0.0606 0.1569 0.0084 0.0252
0.2024 0.8009 0.1651 0.1161
SD 0.4101 0.6600 0.3248 0.2719
Skewness 4.6021 3.5637 5.3228 3.9050
Table 2. Descriptive statistics of general average FTD of A. ludens in 17 wk 2008–2011 in fall at Santa Engracia, Tamaulipas Year
Mean
Median
Variance
2008 2009 2010 2011
0.1223 0.0759 0.0432 0.0325
0.0159 0.0084 0.0000 0.0000
0.0835 0.0953 0.0385 0.0128
SD 0.2496 0.2006 0.1398 0.1028
Skewness 3.9603 5.3427 5.0728 5.0308
Table 3. Indicator semivariogram parameters based on the FTD threshold of 0.1430 in spring from 2008–2011 at Santa Engracia, Tamaulipas Year
Model Threshold Nugget effect
2008 2009 2010 2011
Exp Exp Exp Exp
0.1430 0.1430 0.1430 0.1430
0.0664 0.0240 0.0176 0.0559
Sill 0.1638 0.1470 0.1078 0.1324
Range (m) 750 400 700 850
% 59.5 83.7 83.7 57.8
Exp, exponential; %, ratio between the nugget effect and the sill.
Table 4. Indicator semivariogram parameters based on the FTD threshold of 0.0500 in the fall from 2008–2011 at Santa Engracia, Tamaulipas Year
Model Threshold Nugget effect
2008 2009 2010 2011
Exp Exp Exp Exp
0.0500 0.0500 0.0500 0.0500
0.0492 0.0119 0.0179 0.0215
Sill 0.1507 0.2163 0.0383 0.1573
Range (m) 800 900 900 900
% 67.4 94.5 53.3 86.3
Exp, exponential; %, ratio between the nugget effect and the sill.
east of Santa Engracia, which ranged from 15.2% in spring and 33.1% in the fall. Discussion We observed a high variability of FTD that was likely produced by heterogeneity among adult captures of A. ludens throughout the study area. This heterogeneity affected the spatial distribution of A. ludens infestation over extended citrus areas. However, insect density variability is normal and typically occurs on a scale smaller than the minimum lag distance (Tobin 2004, Garcı´a 2006). Therefore, McPhail traps must be closer to structure a spatial dependence at smaller scale. To achieve that, the trap density should be adjusted according to the population density and commercial season, although in the traditional management of fruit flies in Tamaulipas this would require high costs in human and material resources. Nevertheless, the spatial and temporal management of the pest could be a realistic option for the AW-IPM. Local-scale (orchard) studies have demonstrated that the ranges of effective sampling are between 30 and 40 m in Tephritidae adults (Epsky et al. 2010, Kendra et al. 2010). The flat topography, broad distribution of commercial hosts, and absence of fruiting of the wild host (yellow chapote) in the Sierra Madre Oriental (that could involve movement of populations to the west) led us to assume isotropy. However, anisotropy analyses found no significant deviation from isotropy. Similarly, Plant and Cunningham (1991) indicated that the dispersion of sterile C. capitata flies was not strongly dependent on angular position. Under certain conditions, anisotropy could be present with pronounced altitudinal gradients, as in the case of B. oleae, in Greece, where altitude ranges from 5 to 800 meters above sea level (Castrignano` et al. 2012). The variation of the nugget effect in the exponential models influenced the interpolation, which was a result of—1) High dispersal capacity motivated by the search for resources (Baker and Chan 1991, Hendrichs et al.
6
JOURNAL OF ECONOMIC ENTOMOLOGY
Fig. 3. Probability area distribution based on the reclassification procedure. Models representing probabilities in percentages to exceed the FTD threshold of 0.1430 in the spring at Santa Engracia, Tamaulipas, 2008–2011.
1991) that are more abundant in spring than in fall; A. ludens has a standard distance of typical dispersion of 236–240 m and sometimes up to 9 km (Thomas and Loera-Gallardo 1998, Thomas 2010). 2) High variability among the traps arising from the high population densities, which reduce spatial dependence among sampling points and that has been documented in other insect species (Garcı´a 2006, Tobin 2004). 3) Existence of aggregation or behavior at multiple spatial scales (Liebhold et al. 1993, Petroskii et al. 2014), suggesting the possibility of overlapping populations adapted to a complex agroecological matrix (Aluja and Rull 2009). 4) Anthropogenic influence, such as the application of insecticides that promotes changes to the population distribution. According to Coll (2004), insecticides can affect the stability of populations and facilitate pest outbreaks due to the changes in field distribution. Local-scale studies of sterile flies have indicated that the field distribution of leptokurtosis declines with time and that the fly dispersion ranges are short (Baker and Chan 1991, Plant and Cunningham 1991). Similarly, in this study, a short distance of FTD spatial dependence (range) among sampling points was observed in the semivariograms (Tables 3 and 4). This allows us to
suggest that on large citrus areas the infestation dynamics of A. ludens could be initiated across small patches, and thus addressed through analyses of invasive spatial patterns in short time periods (weekly). Israely et al. (2005) characterized the large-scale population movement of C. capitata (medfly) using GIS to determine its importance for medfly control campaign in Israel. In our probability models, the FTD thresholds of 0.1430 (spring) and 0.0500 (fall) showed differences in the interpolated surface. However, this condition allowed us to identify the spatial distribution of A. ludens infestation according to the time of year, which may be useful in decision making and optimizing regional impact. For example, in C. capitata in Guatemala, a criterion to select priority sites for aerial bait sprays is based on FTD thresholds of 1.0 (Lira 2010). Grapefruit is susceptible to attack by A. ludens, especially in the fall, and a high level of infestation is reported by the fruit flies program. Grapefruit also has a scattered spatial distribution in the citrus areas of Tamaulipas. This is relevant for the interpolated surface because small isolated patches with high probabilities could have resulted from individual traps with high FTD. The grapefruit is considered one of the preferred
2015
VANOYE-ELIGIO ET AL.: SPATIAL ANALYSIS OF A. ludens IN MEXICO
7
Fig. 4. Probability area distribution based on the reclassification procedure. Models representing probabilities in percentages to exceed the FTD threshold of 0.0500 in the fall at Santa Engracia, Tamaulipas, 2008–2011.
commercial hosts of A. ludens (Baker et al. 1944, Robacker and Fraser 2002, Birke et al. 2006). Higher probabilities were predicted in the western regions in both spring and fall, and this was usually characterized by proximity to the Sierra Madre Oriental. The Sierra Madre Oriental is likely rich in the wild host (yellow chapote) of A. ludens (Plummer et al. 1941, Thomas 2003). However, the fruiting period of the yellow chapote in northeastern Mexico during the months of May and June (Thomas 2003) does not coincide with the major citrus production. However, there is evidence to infer movement of populations in yellow chapote to nearby orchards or vice versa (Robacker and Fraser 2002; Thomas 2003, 2012; Quintero et al. 2009). In addition, the wild host is also commonly founded on the Corona riverside that crosses the citrus region. From an ecological point of view, these sites with native vegetation and the neighboring Sierra Madre Oriental are part of agroecosystems with significant benefits in terms of conservation of beneficial organisms (Aluja et al. 2014). However, organophosphate sprays from the fruit flies program may promote changes in pest stability affecting its spatial distribution, as well as the behavior and populations of natural
enemies (Michaud 2003, Coll 2004, Yee and Phillips 2004). Ordano et al. (2013) suggested that chemical action against pest-fly had a detrimental impact on nonpest-fly abundance. The finding presented in this study highlight the importance of the temporal dynamics of the pest and nonpest species populations in the control strategies. Moreover, lower percentages (0–25 and 25–50%) were mainly found in east and at sites distant from the Sierra Madre Oriental. These citrus areas were located mostly adjacent to rangeland and annual crops, which probably induces the population to remain in the orchards and therefore makes them more susceptible to local or regional control activities. However, a characteristic of the area is the formation of large and continuous mosaic of citrus growing areas. This is an advantage for dispersion of fruit flies due to a complex connectivity at different spatial units (landscape, region, or agroecosystem; Aluja and Rull 2009). However, due to the compaction of the citrus growing areas, there is also the possibility for designing an accurate and efficient AW-IPM. The combination of probabilistic models of each season and year contributed to determine homogeneous areas based on risk level (low, medium, and high-risk;
8
JOURNAL OF ECONOMIC ENTOMOLOGY
Fig. 5). This may allow the delineation of subregions and raises the possibility of sampling the closest (spatially) points within certain areas for potential sitespecific management (Tobin 2004, Castrignano` et al. 2012). Medium-risk areas were typically high-risk areas of lower consistency. This can be accounted for by one of two possibilities: 1) areas have high infestation susceptibility because of the invasive and reproductive capacity of A. ludens in adjacent locations with available hosts (Thomas and Loera-Gallardo 1998); and 2) potential population suppression at these sites by the fruit flies program. Based on our models, the sites with better probabilities of success tend to be found in the east of the region. These results suggest a potential site-differential management of low-risk and high-risk sites (Table 5). For example, it might be possible to reduce generalized pesticide use at low-risk sites (Cohen et al. 2008) and to establish plant protection schemes with a high probability of success. We found that the low-risk area accounts for 33.1% (Table 5) of the raster total area in the fall. This finding could be a variable when planning outbreak prevention strategies in January and February. According to Thomas (2003), oviposition in October and November leads the adult population emergence at the beginning of the year. Therefore, activities such as trapping and regional chemical control of the fruit flies program might be more efficient and accurate if the spatial heterogeneity and temporal densities of the pest were considered (Clark et al. 1997, Papadopoulus et al. 2003).
The many low-risk sites identified in this study raise the possibility of integrating the sterile insect technique (SIT) as part of the AW-IPM. Tactical combinations of SIT (i.e., bait spray and male annihilation) have been critical to the eradication of Tephritidae flies from various countries (Suckling et al. 2014). In addition, knowledge advances about diets, hormone use, genetic sexing strain, technology, release strategies, and sexual competitiveness (Utge´s et al. 2011, Go´mez et al. 2013, Orozco et al. 2013, Mubarqui et al. 2014, Morato´ et al. 2015) are likely to improve the prospects for population suppression. The implementation of augmentative biological control at high-risk sites (e.g., at the piedmont of the Sierra Madre Oriental and on the banks of the Corona River) is an environmental-friendly alternative to pesticides. The parasitoids release of the program of fruit flies has been projected to sites with favorable ecological and social conditions (Montoya and Toledo 2010). Therefore, in this case, the objective would be a reduction of fly movement from wild sites to nearby orchards. Diachasmimorpha longicaudata (Hymenoptera: Braconidae) (Ashmead) is a known parasitoid of A. ludens (Ovruski et al. 2000) and is recognized as an attractive form of augmentative biological control (Montoya and Cancino 2004). Finally, the citrus region characterization of Santa Engracia at risk levels with a geostatistical approach integrated into a GIS offered a useful tool to analyze A. ludens populations at a regional scale. This tool combination provided a way to manage high variability of FTD values at the monitoring points. Our finding
Table 5. Values of the final raster classified in an infestation risk level of A. ludens in the spring and in the fall at Santa Engracia, Tamaulipas Spring Total cells 8,846
Fall
Risk
Cell number/ha
%
High Medium Low
1,346 6,541 959
10.8 73.9 15.2
Total cells 8,846
Risk
Cell number/ha
%
High Medium Low
2,932 5,753 157
1.7 65.0 33.1
Fig. 5. Raster final models classified in spring and fall risk levels (low, medium, and high). These models represent the risk of exceeding FTD thresholds of 0.1430 in spring and 0.05 in fall at Santa Engracia, Tamaulipas, from 2008–2011.
2015
VANOYE-ELIGIO ET AL.: SPATIAL ANALYSIS OF A. ludens IN MEXICO
suggests involvement of ecological processes in the west-to-east infestation patterns. In addition, the findings of this study can be used as a baseline for delineating subregions, with a potential for site-differential management in the trapping and control activities of the fruit flies program. Additionally, we propose the future use of both SIT and biological control approaches in the AW-IPM to achieve better long-term suppression of infestation. Acknowledgments We thank the authorities of the Comite´ Estatal de Sanidad Vegetal de Tamaulipas, Mexico, for their support with the trapping database of the National Program Against Fruit Flies in Tamaulipas. The National Council of Science and Technology (CONACYT) supported V. Vanoye for his doctoral studies.
References Cited Alemany, A., M. A. Miranda, R. Alonso, C. M. Escorza. 2006. Changes in the spatial and temporal population density of the Mediterranean fruit fly (Diptera: Tephritidae) in a citrus orchard. Span. J. Agric. Res. 4: 161–166. Aluja, M., and J. Rull. 2009. Managing pestiferous fruit flies (Diptera: Tephritidae) through environmental manipulation, pp. 171–213. In M. Aluja, T. Leskey, and Ch. Vincent, (eds.), Biorational tree fruit pest management. CAB International, Wallingford, United Kingdom. Aluja, M., M. Ordano, L. Guillen, and J. Rull. 2012. Understanding long-term fruit fly (Diptera: Tephritidae) population dynamics: implications for areawide management. J. Econ. Entomol. 105: 823–836. Aluja, M., J. Sivinski, R. Van Driesche, A. Anzures-Dadda, and L. Guille´n. 2014. Pest management through tropical tree conservation. Biodivers. Conserv. 23: 831–853. Baker, M. A. W. E. Stone, C. C. Plummer, and M. McPhail. 1944. A review of studies on the Mexican fruit fly and related Mexican species. United States Department of Agriculture Miscellaneous Publication 531, Washington D.C. Baker, P. S. and A. S. Chan. 1991. Appetitive dispersal of sterile fruit flies: aspects of the methodology and analysis of trapping studies. J. Appl. Entomol. 112: 263–273. Barabas, N., P. Goovaerts, and P. Adriaens. 2001. Geostatistical assessment and validation of uncertainty for threedimensional dioxin data from sediments in an estuarine river. Environ. Sci. Technol. 35: 3294–3301. Birke, A., M. Aluja, P. Greany, E. Bigurra, D. Pe´rezStaples, and R. McDonald. 2006. Long aculeus and behavior of Anastrepha ludens render gibberellic acid ineffective as an agent to reduce ‘ruby red’ grapefruit susceptibility to the attack of this pestiferous fruit fly in commercial groves. J. Econ. Entomol. 99: 1184–1193. Brenner, R. J., D. A. Focks, R. T. Arbogast, D. K. Weaver, and D. Shuman, 1998. Practical use of spatial analysis in precision targeting for integrated pest management. Am. Entomol. 44: 79–102. Castrignano`, A., L. Boccaccio, Y. Cohen, D. Nestel, I. Kounatidis, N. T. Papadopoulos, D. De Benedetto, and P. Mavragani-Tsipidou. 2012. Spatio-temporal population dynamics and area-wide delineation of Bactrocera oleae monitoring zones using multi-variate geostatistics. Precis. Agric. 13: 421–441. Cohen, Y., A. Cohen, A. Hetzroni, V. Alchanatis, D. Broday, Y. Gazit, and D. Timar. 2008. Spatial decision support system for medfly control in citrus. Comput. Electron. Agric. 62: 107–117.
9
Coll, M. 2004. Precision agriculture approaches in support of ecological engineering for pest management, pp. 133–142. In G. M. Gurr, S. D. Wratten, and M. A. Altieri (eds.), Ecological ingeneering for pest management. Advances in habitat manipulation for arthropods. Sciro, Australia. Deutsch, C. V., and A. G. Journel. 1992. GSLIB: Geostatistical Software library and user´s guide. Oxford University Press, New York, NY. Epsky, N. D., H. R. Espinoza, P. E. Kendra, R. Abernathy, D. Midgarden, and R. R. Heath. 2010. Effective sampling range of a synthetic protein-based attractant for Ceratitis capitata (Diptera: Tephritidae). J. Econ. Entomol. 103: 1886–1895. Eydel, J., R. Ları´n, and E. Garea. 2011. Hacia me´todos de ana´lisis de datos espaciales raster en el nivel sema´ntico. Comput. Sist. 15: 91–106. Fleischer, S. J., P. E. Blom, and R. Weisz. 1999. Sampling in precision IPM: When the objective is a map. Phytopathology 89: 1112–1118. Garcı´a, F.J.M. 2006. Analysis of the spatio-temporal distribution of Helicoverpa armigera Hb. in a tomato field using stochastic approach. Biosyst. Eng. 93: 253–259. Go´mez, Y., P.E.A. Teal, and R. Pereira. 2013. Enhancing efficacy of Mexican fruit fly SIT programmes by large-scale incorporation of methoprene into pre-release diet. J. App. Entomol. 137: 252–259. Goovaerts, P. 1997. Geostatistics for natural resources evaluation. Applied geostatistics series. Oxford University Press, New York. Grunwald, S., P. Goovaerts, C. M. Bliss, N. B. Comerford, and S. Lamsal. 2006. Incorporation of auxiliary information in the geostatistical simulation of soil nitrate nitrogen. Vadose Zone J. 5: 391–404. Guidotti, D., G. Ragaglini, and R. Petacchi. 2005. Analysis of spatio-temporal Bactrocera oleae (Diptera, Tephritidae) infestation distributions obtained from a large-scale monitoring network and its importance to IPM. IOBC WPRS Bull. 28: 13–18. Gutie´rrez, J. M. 2010. El Programa de Moscas de la Fruta en Me´xico, pp. 3–10. In P. Montoya, J. Toledo and E. Herna´ndez (eds.), Moscas de la fruta: Fundamentos y Procedimientos para su Manejo. S y G, Me´xico, D.F. Hendrichs, J., B. I. Katsoyannos, D. R. Papaj, and R. J. Procopy. 1991. Sex differences in movement between natural feeding and mating sites and tradeoffs between food consumption, mating success and predator evasion in Mediterranean fruit flies (Diptera: Tephritidae). Oecologia 86: 223–231. Hendrichs, J., P. Kenmore, A. S Robinson, and M.J.B. Vreysen. 2007. Area-wide integrated pest management (AWIPM): Principles, practice and prospects, pp. 3–33. In M.J.B. Vreysen, A. S Robinson and J. Hendrichs (eds.), AreaWide control of insect pests: from research to field implementation. Springer, Dordrecht, The Netherlands. Hernandez-Ortiz, V., and M. Aluja. 1993. Listado de especies del ge´nero neotropical Anastrepha (Diptera: Tephritidae) con notas sobre su distribucio´n y plantas hospederas. Folia Entomol. Mex. 88:89–105. (IAEA) International Atomic Energy Agency. 2003. Trapping guidelines for areawide fruit programmes. Insect Pest Control Section, International Atomic Energy Agency, Vienna, Austria. Isaaks, E. H., and R. M. Srivastava. 1989. An introduction to applied geostatistics. Oxford University Press. New York, NY. Israely, N., Y. Ziv, and R. Galun. 2005. Metapopulation spatial-temporal distribution patterns of Mediterranean fruit fly (Diptera: Tephritidae) in a patchy environment. Ann. Entomol. Soc. Am. 98: 302–308. Journel, A. G. 1983. Non parametric estimation on spatial distributions. Math. Geol. 15: 445–468. Kendra, E. P., D. N. Epsky, and R. R. Heath. 2010. Effective sampling range of food-based attractants for female
10
JOURNAL OF ECONOMIC ENTOMOLOGY
Anastrepha suspensa (Diptera: Tephritidae). J. Econ. Entomol. 103: 533–540. Liebhold, A. M., E. R. Rossi, and P. W. Kemp. 1993. Geostatistics and geographic information systems in applied insect ecology. Annu. Rev. Entomol. 38: 303–327. Liebhold, A. M., and A. A. Sharov. 1998. Testing for correlation in the presence of spatial autocorrelation in insect count data, pp. 11–117. In P. Baumgartner, P. Brandmayr, and B.F.J. Manly (eds.), Population and community ecology for insect management and conservation. Balkema, Rotterdam. Lira, E. 2010. Uso de Sistemas de Informacio´n Geogra´fica, pp. 159–166. In P. Montoya, J. Toledo, E. Herna´ndez. (eds.), Moscas de la Fruta: Fundamentos y Procedimientos para su Manejo. S y G, Me´xico, D.F. Lyon, S. W., A. J. Lembo Jr, M. T. Walter, and T. S. Steenhuis. 2006. Defining probability of saturation with indicator kriging on hard and soft data. Adv. Water Resour. 29: 181–196. Mangan, R. L., E. R. Frampton, D. B. Thomas, and D. S. Moreno. 1997. Application of the maximum pest limit concept to quarantine security standards for the Mexican fruit fly (Diptera: Tephritidae). J. Econ. Entomol. 90: 1433–1440. Meats, A. 2007. Dispersion of fruit flies (Diptera: Tephritidae) at high and low densities and consequences of mismatching dispersions of wild and sterile flies. Fla. Entomol. 90: 136–146. Michaud, J. P. 2003. Toxicity of fruit fly baits to beneficial insects in citrus. J. Insect Sci. 3: 1–9. Montoya, P. and J. Cancino. 2004. Control biolo´gico por aumento en moscas de la fruta (Diptera: Tephritidae). Folia Entomol. Mex. 43: 257–270. Montoya, P. and J. Toledo. 2010. Estrategias de control biolo´gico, pp. 169–182. In P. Montoya, J. Toledo, E. Herna´ndez. (eds.), Moscas de la Fruta: Fundamentos y Procedimientos para su Manejo. S y G, Me´xico, D.F. Morato´, S., T. Shelly, J. Rull, and M. Aluja. 2015. Sexual competitiveness of Anastrepha ludens (Diptera: Tephritidae) males exposed to Citrus aurantium and Citrus paradisi essential oils. J. Econ. Entomol. 108: 621–628. Mubarqui, R. L., R. C. Perez, R. A. Kladt, J.L.Z. Lopez, A. Parker, M. T. Seck, B. Sall, and J. Bouyer. 2014. The smart aerial release machine, a universal system for applying the sterile insect technique. PLoS ONE 9: e103077. Ovruski, S., M. Aluja, J. Sivinski, and R. Wharton. 2000. Hymenopteran parasitoids on fruit-infesting Tephritidae (Diptera) in Latin America and the southern United States: Diversity, distribution, taxonomic status and their use in fruit fly biological control. Integrated Pest. Manage. Rev. 5: 81–107. Ordano, M., L. Guille´n, J. Rull, R. Lasa, and M. Aluja. 2013. Temporal dynamics of diversity in a tropical fruit fly (Tephritidae) ensemble and their implications on pest management and biodiversity conservation. Biodivers. Conserv. 22: 1557–1575. Orozco, D., J. S. Meza, S. Zepeda, E. Solı´s, and J. L. Quintero-Fong. 2013. Tapachula-7, a new genetic sexing strain of the Mexican fruit fly (Diptera: Tephritidae): Sexual compatibility and competitiveness. J. Econ. Entomol. 106: 735–741. Papadopoulus, N. T., B. I. Katsoyannos, and D. Nesstel. 2003. Spatial autocorrelation of a Ceratitis capitata (Diptera: Tephritidae) adult population in a mixed deciduous fruit orchard in Northern Greece. Environ. Entomol. 32: 319–326. Petroskii, S., N. Petrovskayab, and D. Bearup. 2014. Multiscale approach to pest insect monitoring: Random walks,
pattern formation, synchronization, and networks. Phys. Life Rev. 11: 467–525. Plant, R. E. and R. T. Cunningham. 1991. Analyses of the dispersal of sterile Mediterranean fruit flies (Diptera: Tephritidae). Environ. Entomol. 20: 1493–1503. Plummer, C. C., M. McPhail, and W. J. Monk. 1941. The yellow chapote, a native host of the Mexican fruit fly. USDA. Tech. Bull. 775: 51. Quintero, V. P., J.I.L. Arroyo, J. L. Gallardo, J. Rull, E. R. Robles, E. C. Mondaca, S. H Delgado, N. M Perez, and M. A. Schuneman. 2009. Genetic differences between Anastrepha ludens (Loew) populations stemming from a native and an exotic host in NE Mexico. Agric. Te´cnica en Me´x. 35: 323–331. Robacker, D. C. and I. Fraser. 2002. Do Mexican fruit flies (Diptera: Tephritidae) prefer grapefruit to yellow chapote, a native host?. Fla. Entomol. 85: 481–487. (SAGARPA) Secretarı´a de Agricultura, Ganaderı´a, Desarrollo Rural, Pesca y Alimentacio´n .2013. Padro´n de productores de cı´tricos georeferenciados 2010-2012 Tamaulipas. Servicio de Informacio´n Agroalimentario y Pesquera, Gobierno del estado de Tamaulipas, Me´xico, D.F. Scalon, J. D., G. de F. Alves, M.B.L. Avelar, and M. S. Zacarias. 2011. Spatial distribution of the coffee leaf-miner (Leucoptera coffeella (Gue´rin-Me`neville & Perrottet, 1842) in an organic coffee (Coffea arabica L.) field in formation. Coffee Sci. 6: 226–232. Suckling, D. M., J. M. Kean, L. D. Stringer, C. Ca´ceresBarrios, J. Hendrichs, J. Reyes-Flores, and B. C. Dominiak. 2014. Eradication of tephritid fruit fly pest populations: outcomes and prospects. Pest Manage. Sci. doi: 10.1002/ ps.3905. Steyskal, G. C. 1977. History and use of the McPhail trap. Fla. Entomol. 60: 11–16. Tscharntke, T., I. Steffan-Dewenter, A. Kruess, and C. Thies. 2002. Characteristics of insect populations on habitat fragments: a mini review. Ecol. Res. 17: 229–239. Thomas, D. B. 2003. Reproductive phenology of the Mexican fruit fly, Anastrepha ludens (Loew) (Diptera: Tephritidae) in the Sierra Madre Oriental, northern Mexico. Neotrop. Entomol. 32: 385–397. Thomas, D. B. 2010. Dispersio´n y dina´mica poblacional, pp. 107–112. In: P. Montoya, J. Toledo and E. Herna´ndez (eds.), Moscas de la fruta: Fundamentos y Procedimientos para su Manejo. S y G, Me´xico, D.F. Thomas, D. B. 2012. Mexican Fruit Fly (Diptera: Tephritidae) and the phenology of its native host plant yellow chapote (Rutaceae) in Mexico. J. Entomol. Sci. 47: 1–16. Thomas, D., and J. Loera-Gallardo, 1998. Dispersal and longevity of mass-released, sterilized Mexican fruit flies (Diptera: Tephritidae). Environ. Entomol. 27: 1045–1052. Tobin, P. C. 2004. Estimation of the spatial autocorrelation function: consequences of sampling dynamic populations in space and time. Ecography 27: 767–775. Utge´s, M. E., J. C. Vilardi, A. Oropeza, J. Toledo, and P. Liedo. 2011. Pre-release diet effect on field survival and dispersal of Anastrepha ludens and Anastrepha obliqua (Diptera: Tephritidae). J. App. Entomol. 137: 163–177. Yee, W. L., and A. Phillips. 2004. Differential mortality of natural enemies exposed to avocado leaves treated with malathion bait spray during a Mediterranean fruit fly eradication program. Southwest. Entomol. 29: 175–184. Received 18 December 2014; accepted 4 May 2015.
